CN109408998A - Estimating method for fatigue life is carried out based on sample incremental quick obtaining stress spectra - Google Patents
Estimating method for fatigue life is carried out based on sample incremental quick obtaining stress spectra Download PDFInfo
- Publication number
- CN109408998A CN109408998A CN201811324437.5A CN201811324437A CN109408998A CN 109408998 A CN109408998 A CN 109408998A CN 201811324437 A CN201811324437 A CN 201811324437A CN 109408998 A CN109408998 A CN 109408998A
- Authority
- CN
- China
- Prior art keywords
- data
- stress
- sample
- crane
- radial basis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
Abstract
The invention discloses one kind to carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra, the following steps are included: extract real-time crane hoisting load capacity and small truck position sample data as input from overhead traveling crane data recorder, using trained radial basis neural network, it can be obtained sample data and continue to increase the lower corresponding equivalent stress value of bridge crane any position point, to draw m- stress curve when this;According to the when m- stress curve of drafting, utilize rain flow method, obtain include mean value and amplitude stress spectra;Obtained stress spectra is extracted, using based on damage tolerance --- fracture mechanics method calculates the fatigue surplus life of bridging crane main beam.The present invention is for predicting in real time and assessing service life and the reliability of bridging crane main beam.The present invention solve the problems, such as current general-purpose overhead crane in face of sample data incremental can not quick obtaining when m- stress curve obtain stress spectra and carry out analysis of Fatigue-life.
Description
Technical field
The invention belongs to crane estimation of fatigue life fields, and in particular to one kind is based on sample incremental quick obtaining
Stress spectra carries out estimating method for fatigue life, is also applied for the estimation of fatigue life of other types crane.
Background technique
Bridge crane carries out the lifting equipment of product handling as workshop, and work characteristics is to do intermittent exercise, i.e.,
The corresponding mechanism of the movements such as feeding, migration, unloading is to work alternatively in a working cycles, and crane is sent out in the market
It opens up and using more and more extensive.With increasing for bridge crane usage quantity, crane using the safety accident caused not yet
It is disconnected to rise, and once accident occurs, social influence is severe, brings huge loss to the country and people.According to statistics, China is every
Year because caused by hoisting machinery casualty accident account for 15% or so of the generation of current year accident, and with the development of economy, people for
The performance of bridge crane has higher requirement, and the indexs such as big elevating capacity, high frequency operation are brought into schedule, these rigid indexs
So that bridge crane in-service period more " busy ", has undoubtedly aggravated the accident frequency of bridge crane.Overhead traveling crane as a result,
The safety problem of machine has become the important topic faced in country and industry.
When in the prior art to Structural Metallic Fatigue remaining life fail-safe analysis, the characteristic parameter by crane is needed
Record m- stress curve when obtaining, and stress spectra is calculated.But it at present can not be to the real-time sample data incremental of crane
It is analyzed, most of calculate is analyzed subsequently through to the when m- stress curve in a period of time, to obtain knot
By.Therefore on the problem of carrying out real-time dynamic forecast for crane and assess its remaining life, at present when m- stress curve obtain
Method is taken to have certain limitation.
Summary of the invention
In order to solve current general-purpose overhead crane in face of sample data incremental can not quick obtaining when m- stress
Curve obtains stress spectra and carries out analysis of Fatigue-life problem, and provides a kind of fast and easy acquisition time-stress curve, effectively meter
That calculates fatigue life carries out estimating method for fatigue life based on sample incremental quick obtaining stress spectra.
Used technical solution is the present invention to achieve the goals above:
One kind carrying out estimating method for fatigue life based on sample incremental quick obtaining stress spectra, comprising the following steps: from bridge
Extract real-time crane hoisting load capacity and small truck position sample data as input in formula lifting data recorder use
Trained radial basis neural network can be obtained sample data and continue to increase the lower bridge crane any position point pair
The equivalent stress value answered, to draw m- stress curve when this;According to the when m- stress curve of drafting, rain-flow counting is utilized
Method obtains the stress spectra comprising mean value and amplitude;Obtained stress spectra is extracted, using based on damage tolerance --- fracture mechanics method
Calculate the fatigue surplus life of bridging crane main beam.The present invention utilizes training by the sample of extract real-time bridge crane
Good radial base neural net can carry out crane any position point real-time time-in face of crane sample incremental and answer
Force curve is drawn, and obtains stress spectra, and estimate the fatigue surplus life of bridging crane main beam, for prediction in real time and assessment bridge
The service life of formula crane girder and reliability.
Further, the construction step of the radial basis neural network is as follows: establishing overhead traveling crane by ANSYS
Machine simulation model obtains learning sample data, and constructs radial basis neural network;According to obtained radial base neural net
Model and learning sample data, selection 1-2 group data are test data, remaining group data is learning sample, utilize radial base letter
Several pairs of radial base neural nets are trained, and export the equivalent of learning sample by trained radial basis neural network
Stress value and ANSYS, which analyze to obtain the location point equivalent stress value, to be compared, and confirms trained radial base neural net mould
The accuracy of type obtains the radial basis neural network of optimal model parameters.
Further, the acquisition methods of the learning sample are as follows: according to bridge crane hoisting heavy trolley from a left side
End runs to the different lifted loads of right end and small truck position is the input quantity of learning sample, and it is tired to calculate such bridge crane
The equivalent stress value that the corresponding ANSYS of labor key point is calculated is the output quantity of learning sample, obtains learning sample data.
Preferably, the radial basis neural network be before be made of input layer, hidden layer and output layer three layers to
Type network, for hidden layer using radial basis function as excitation function, which is generally Gaussian function.The radial direction
Base neural net model is before the one kind constructed based on function approaches theory to type network.Radial Basis Function Neural is
One network only having levels.In middle layer, it is to replace swashing for traditional global response to the radial basis function of layout response
It sends a letter number.Input layer number is 2, and output layer number of nodes is 1.
Further, the step of rain flow method is as follows:
1) it pre-processes
The first m- stress curve data of clock synchronization are modified, and are made to obtain data and are more favorable for record recurring number;
2) data point reuse
Determine whether data are even number, if not even number, then remove the last one data, after making data processing when m- answer
Force curve data are even number, form totally-enclosed counter model;
3) data are docked
Data in step 2 are adjusted, starting point top or lowest trough are made, then docks in head and the tail, is carrying out in this way
One time total data can be completed in rain-flow counting;
4) cycle count
Data after docking are subjected to cycle count, take out all circulations, record includes the stress spectra of amplitude and mean value.
Further, the amendment data include: data compression, detection peak-to-valley value, the invalid stress amplitude of removal.
Preferably, data compression is that consecutive identical data are carried out with compression to remove identical data, only retains one.
Preferably, the detection peak-to-valley value is to extract peak value in data and valley, Rule of judgment are as follows: for
I data element is worked asWhen, value is not that peak value is exactly valley.
Preferably, the invalid stress amplitude of removal is to remove the small stress amplitude of some pairs of aging effects.
Further, based on damage tolerance --- fracture mechanics method estimates the fatigue surplus life of crane.
The present invention utilizes trained radial base neural net, can quick obtaining crane any position point by computer
Real-time time-stress curve obtain stress spectra and estimate the fatigue surplus life of its girder, so that it is existing to greatly save crane
Actual measurement, or according to raising process the ANSYS complicated processes analyzed and investment, realize real-time acquisition time-stress curve and
The purpose of accurate estimation crane girder fatigue surplus life;Based on trained radial base neural net, establish to such
Real-time time-stress curve acquisition of type bridge crane, without will after work according to lifted load again into
The drafting of m- stress curve when row has and realizes simple, effectively quick and can predict and assess in real time bridging crane main beam
The characteristics of service life and reliability.
Detailed description of the invention
Fig. 1 is flow diagram of the invention;
Fig. 2 is radial base neural net structural schematic diagram of the present invention;
Fig. 3 is one course of work of crane by the ANSYS position-stress curve analyzed and by radial base nerve net
Position-stress diagrams that network obtains;
Fig. 4 is ANSYS analysis and neural network error curve.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.
Embodiment 1
As shown in Figures 1 and 2, Fatigue Life Assessment is carried out based on sample incremental quick obtaining stress spectra in the present embodiment
Method, comprising the following steps: extract real-time crane hoisting load capacity and small truck position from overhead traveling crane data recorder
Sample data as input can be obtained under sample data continues to increase using trained radial basis neural network
The corresponding equivalent stress value of bridge crane any position point, thus m- stress curve when drawing;According to drafting when it is m-
Stress curve obtains the stress spectra comprising mean value and amplitude using rain flow method;Obtained stress spectra is extracted, using being based on
Damage tolerance --- fracture mechanics method calculates the fatigue surplus life of bridging crane main beam.The present invention passes through extract real-time bridge
The sample of formula crane utilizes trained radial base neural net, can carry out any position in face of crane sample incremental
It sets a real-time time-stress curve to draw, obtains stress spectra, and estimate the fatigue surplus life of bridging crane main beam, be used for
The service life and reliability of prediction in real time and assessment bridging crane main beam.
Further, the construction step of the radial basis neural network is as follows: establishing overhead traveling crane by ANSYS
Machine simulation model obtains learning sample data, and constructs radial basis neural network;According to obtained radial base neural net
Model and learning sample data, choosing 1 group of data is test data, remaining group data is learning sample, utilizes radial basis function
Radial base neural net is trained, and by trained radial basis neural network export learning sample etc. effects
Force value and ANSYS, which analyze to obtain the location point equivalent stress value, to be compared, and confirms trained radial basis neural network
Accuracy, obtain the radial basis neural network of optimal model parameters.
Further, the acquisition methods of the learning sample are as follows: according to bridge crane hoisting heavy trolley from a left side
End runs to the different lifted loads of right end and small truck position is the input quantity of learning sample, calculates such bridge crane and appoints
The equivalent stress value that the corresponding ANSYS of location point is calculated of anticipating is the output quantity of learning sample, obtains learning sample data.?
Practise sample data be bridge crane be under normal operating conditions some typical lifted loads and small truck position and its it is right
(lifted load (lifting capacity) is respectively 0,5t, 15t, 25t, 35t, 45t, 55t, 65t, 75t to the equivalent stress value answered;Small parking stall
Set from 3m and be moved to 25m), to establish a database model, learning sample is made to meet the actual use operating condition of crane.
Preferably, the radial basis neural network be before be made of input layer, hidden layer and output layer three layers to
Type network, for hidden layer using radial basis function as excitation function, which is generally Gaussian function.The radial direction
Base neural net model is before the one kind constructed based on function approaches theory to type network.Radial Basis Function Neural is
One network only having levels.In middle layer, it is to replace swashing for traditional global response to the radial basis function of layout response
It sends a letter number.Input layer number is 2, and output layer number of nodes is 1.
The weight vector w1 that each neuron of hidden layer is connected with input layeriWith input vector vector Xq(q-th of expression defeated
Incoming vector) the distance between be multiplied by threshold values b1iAs the input of itself, it can thus be concluded that the input of i-th of neuron of hidden layer
For, export and be.The adjustable function of threshold values b1 of radial basis function
Sensitivity, but be more often known as spread function with another parameter C(in real work).In MATLAB Neural Network Toolbox,
The relationship of b1 and C is, the output of hidden layer neuron becomes at this timeThe input of output layer is each hidden layer mind
Weighted sum through member output.Therefore it exports and is。
The training process of radial basis neural network is divided into two steps: the first step learns for no tutor's formula, determines that training is defeated
Enter the weight w 1 of layer Yu implicit interlayer;Second step is to have the study of tutor's formula, determines the weight w 2 of training hidden layer and output layer.?
Before training, it is desirable to provide the extension constant C of input vector X, corresponding output vector T and radial basis function.Trained target
It is the final weight w 1 for seeking two layers, w2 and threshold values b1, b2(when hidden layer unit number is equal to input vector, takes b2=0).
A radial basis function network is designed with Newrbe function now.Its call format be net=newrbe (P, T, goal,
spread,MN,DF);Wherein parameter P is that P × Q of Q group input quantity composition ties up matrix;Parameter T is S × Q of Q group output quantity composition
Tie up matrix;Parameter goal is the expansion rate of radial basis function;Parameter spread is the distribution of radial basis function;Parameter MN is mind
Maximum number through member, is defaulted as Q;Parameter DF is added neuron number between display twice.Output parameter net is to return
Return value, a radial basis neural network.
Further, the step of rain flow method is as follows:
1) it pre-processes
The first m- stress curve data of clock synchronization are modified, and are made to obtain data and are more favorable for record recurring number;
2) data point reuse
Determine whether data are even number, if not even number, then remove the last one data, after making data processing when m- answer
Force curve data are even number, form totally-enclosed counter model;
3) data are docked
Data in step 2 are adjusted, starting point top or lowest trough are made, then docks in head and the tail, is carrying out in this way
One time total data can be completed in rain-flow counting;
4) cycle count
Data after docking are subjected to cycle count, take out all circulations, record includes the stress spectra of amplitude and mean value.
Further, the amendment data include: data compression, detection peak-to-valley value, the invalid stress amplitude of removal.
Preferably, data compression is that consecutive identical data are carried out with compression to remove identical data, only retains one.
Preferably, the detection peak-to-valley value is to extract peak value in data and valley, Rule of judgment are as follows: for
I data element is worked asWhen, value is not that peak value is exactly valley.
Preferably, the invalid stress amplitude of removal shows the small stress amplitude of some pairs of aging effects of removal.
Further, based on damage tolerance --- fracture mechanics method estimates the fatigue surplus life of crane.
Embodiment 2
As shown in Figures 1 and 2, Fatigue Life Assessment is carried out based on sample incremental quick obtaining stress spectra in the present embodiment
Method, comprising the following steps: extract real-time crane hoisting load capacity and small truck position from overhead traveling crane data recorder
Sample data as input can be obtained under sample data continues to increase using trained radial basis neural network
The corresponding equivalent stress value of bridge crane any position point, thus m- stress curve when drawing;According to drafting when it is m-
Stress curve obtains the stress spectra comprising mean value and amplitude using rain flow method;Obtained stress spectra is extracted, using being based on
Damage tolerance --- fracture mechanics method calculates the fatigue surplus life of bridging crane main beam.The present invention passes through extract real-time bridge
The sample of formula crane utilizes trained radial base neural net, can carry out any position in face of crane sample incremental
It sets a real-time time-stress curve to draw, obtains stress spectra, and estimate the fatigue surplus life of bridging crane main beam, be used for
The service life and reliability of prediction in real time and assessment bridging crane main beam.
Further, the construction step of the radial basis neural network is as follows: establishing overhead traveling crane by ANSYS
Machine simulation model obtains learning sample data, and constructs radial basis neural network;According to obtained radial base neural net
Model and learning sample data, choosing 2 groups of data is test data, remaining group data is learning sample, utilizes radial basis function
Radial base neural net is trained, and by trained radial basis neural network export learning sample etc. effects
Force value and ANSYS, which analyze to obtain the location point equivalent stress value, to be compared, and confirms trained radial basis neural network
Accuracy, obtain the radial basis neural network of optimal model parameters.
Further, the acquisition methods of the learning sample are as follows: according to bridge crane hoisting heavy trolley from a left side
End runs to the different lifted loads of right end and small truck position is the input quantity of learning sample, calculates such bridge crane and appoints
The equivalent stress value that the corresponding ANSYS of location point is calculated of anticipating is the output quantity of learning sample, obtains learning sample data.?
Practise sample data be bridge crane be under normal operating conditions some typical lifted loads and small truck position and its it is right
(lifted load (lifting capacity) is respectively 0,5t, 15t, 25t, 35t, 45t, 55t, 65t, 75t to the equivalent stress value answered;Small parking stall
Set from 3m and be moved to 25m), to establish a database model, learning sample is made to meet the actual use operating condition of crane.
Preferably, the radial basis neural network be before be made of input layer, hidden layer and output layer three layers to
Type network, for hidden layer using radial basis function as excitation function, which is generally Gaussian function.The radial direction
Base neural net model is before the one kind constructed based on function approaches theory to type network.Radial Basis Function Neural is
One network only having levels.In middle layer, it is to replace swashing for traditional global response to the radial basis function of layout response
It sends a letter number.Input layer number is 2, and output layer number of nodes is 1.
The weight vector w1 that each neuron of hidden layer is connected with input layeriWith input vector vector Xq(q-th of expression defeated
Incoming vector) the distance between be multiplied by threshold values b1iAs the input of itself, it can thus be concluded that the input of i-th of neuron of hidden layer
For, export and be。
The sensitivity of the adjustable function of threshold values b1 of radial basis function, but be more often known as extending letter with another parameter C(in real work
Number).In MATLAB Neural Network Toolbox, the relationship of b1 and C are, hidden layer neuron is defeated at this time
Become outThe input of output layer is each hidden
The weighted sum of the output of neuron containing layer.Therefore it exports and is。
The training process of radial basis neural network is divided into two steps: the first step learns for no tutor's formula, determines that training is defeated
Enter the weight w 1 of layer Yu implicit interlayer;Second step is to have the study of tutor's formula, determines the weight w 2 of training hidden layer and output layer.?
Before training, it is desirable to provide the extension constant C of input vector X, corresponding output vector T and radial basis function.Trained target
It is the final weight w 1 for seeking two layers, w2 and threshold values b1, b2(when hidden layer unit number is equal to input vector, takes b2=0).
A radial basis function network is designed with Newrbe function now.Its call format be net=newrbe (P, T, goal,
spread,MN,DF);Wherein parameter P is that P × Q of Q group input quantity composition ties up matrix;Parameter T is S × Q of Q group output quantity composition
Tie up matrix;Parameter goal is the expansion rate of radial basis function;Parameter spread is the distribution of radial basis function;Parameter MN is mind
Maximum number through member, is defaulted as Q;Parameter DF is added neuron number between display twice.Output parameter net is to return
Return value, a radial basis neural network.
Further, the step of rain flow method is as follows:
1) it pre-processes
The first m- stress curve data of clock synchronization are modified, and are made to obtain data and are more favorable for record recurring number;
2) data point reuse
Determine whether data are even number, if not even number, then remove the last one data, after making data processing when m- answer
Force curve data are even number, form totally-enclosed counter model;
3) data are docked
Data in step 2 are adjusted, starting point top or lowest trough are made, then docks in head and the tail, is carrying out in this way
One time total data can be completed in rain-flow counting;
4) cycle count
Data after docking are subjected to cycle count, take out all circulations, record includes the stress spectra of amplitude and mean value.
Further, the amendment data include: data compression, detection peak-to-valley value, the invalid stress amplitude of removal.
Preferably, data compression is that consecutive identical data are carried out with compression to remove identical data, only retains one.
Preferably, the detection peak-to-valley value is to extract peak value in data and valley, Rule of judgment are as follows: for
I data element is worked asWhen, value is not that peak value is exactly valley.
Preferably, the invalid stress amplitude of removal shows the small stress amplitude of some pairs of aging effects of removal.
Further, based on damage tolerance --- fracture mechanics method estimates the fatigue surplus life of crane.
The present embodiment is corresponding to be somebody's turn to do using bridge crane lifted load (lifting capacity) and small truck position as the input of learning sample
The equivalent stress value of location point is that the hope of learning sample exports, m- stress curve when obtaining learning sample, and drawing;This reality
Applying the learning sample data in example is that bridge crane simulation model is established by ANSYS, passes through bridge crane hoisting heavy
It is learning sample data that trolley, which runs to 25m with the equivalent stress value of corresponding tired key point from 3m,.Table 1 is to lifted load
(lifting capacity) is respectively 0,5t, 15t, 25t, 35t, 45t, 55t, 65t, 75t, small truck position from 3m run to 25m with it is corresponding tired
The part sample data of labor key point equivalent stress value.
1 bridge crane lifted load of table, small truck position and tired key point equivalent stress value part sample data
(3) train RBF Neural Network model
Neural network after only training just is able to achieve its function.The learning sample normalization that step (2) is obtained, training is by step
Suddenly the radial base neural net that (1) determines, make neural network to the error very little of the output of non-learning sample and desired value until
Meet the requirement of application.
(4) when bridge crane m- stress curve acquisition
As shown in Figs. 3-4, step (3) trained radial basis neural network establishes different lifted loads, small parking stall
It sets and the relationship of the equivalent stress of corresponding tired key point.It is such as 70t, 50t by lifted load, from lifting position 5m to unloading
Position 15m is input to step (3) trained radial basis neural network, and the when m- stress that can obtain the crane is bent
Line, result table 2, it can be seen that it and input sample are consistent, and confirming that trained radial basis neural network is can
Capable.
The tired key point equivalent stress of 2 bridge crane lifted load 75t of table
(5) fatigue surplus life is estimated
By the when m- stress curve for the bridge crane that radial basis neural network obtains, extracted using rain flow method
The cycle-index and global cycle number of stress spectra including amplitudes at different levels, mean value, according to based on damage tolerance --- fracture mechanics
Method calculates the fatigue surplus life of crane girder.
Claims (9)
1. one kind carries out estimating method for fatigue life based on sample incremental quick obtaining stress spectra, which is characterized in that including
Following steps: extract real-time crane hoisting load capacity and small truck position are as input from overhead traveling crane data recorder
Sample data can be obtained sample data and continued to increase the lower overhead traveling crane using trained radial basis neural network
The corresponding equivalent stress value of machine any position point, to draw m- stress curve when this;It is bent according to the when m- stress of drafting
Line obtains the stress spectra comprising mean value and amplitude using rain flow method;Obtained stress spectra is extracted, is held using based on damage
Limit --- fracture mechanics method calculates the fatigue surplus life of bridging crane main beam.
2. according to claim 1 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the construction step of the radial basis neural network is as follows: establishing bridge crane emulation by ANSYS
Model obtains learning sample data, and constructs radial basis neural network;According to obtained radial basis neural network and
Learning sample data, selection 1-2 group data are test data, remaining group data is learning sample, utilize radial basis function diameter
It is trained to base neural net, and exports the equivalent stress value of learning sample by trained radial basis neural network
It analyzes to obtain the location point equivalent stress value with ANSYS and compare, confirm the standard of trained radial basis neural network
True property, obtains the radial basis neural network of optimal model parameters.
3. according to claim 2 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the acquisition methods of the learning sample are as follows: being run to according to bridge crane hoisting heavy trolley from left end
The different lifted loads of right end and small truck position are the input quantity of learning sample, calculate such bridge crane fatigue key point
The equivalent stress value that corresponding ANSYS is calculated is the output quantity of learning sample, obtains learning sample data.
4. according to any one of claim 1 to 2 carry out the tired longevity based on sample incremental quick obtaining stress spectra
Order appraisal procedure, which is characterized in that the radial basis neural network is three be made of input layer, hidden layer and output layer
To type network before layer, hidden layer is using radial basis function as excitation function.
5. according to claim 1 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the step of rain flow method, is as follows:
1) it pre-processes
The first m- stress curve data of clock synchronization are modified, and are made to obtain data and are more favorable for record recurring number;
2) data point reuse
Determine whether data are even number, if not even number, then remove the last one data, after making data processing when m- answer
Force curve data are even number, form totally-enclosed counter model;
3) data are docked
Data in step 2 are adjusted, starting point top or lowest trough are made, then docks in head and the tail, is carrying out in this way
One time total data can be completed in rain-flow counting;
4) cycle count
Data after docking are subjected to cycle count, take out all circulations, record includes the stress spectra of amplitude and mean value.
6. according to claim 5 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the amendment data include: data compression, detection peak-to-valley value, the invalid stress amplitude of removal.
7. according to claim 6 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the data compression is that consecutive identical data are carried out with compression to remove identical data, only retain one.
8. according to claim 6 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the detection peak-to-valley value is to extract peak value in data and valley, Rule of judgment are as follows: for i-th
Data element is worked asWhen, value is not that peak value is exactly valley.
9. according to claim 6 carry out estimating method for fatigue life based on sample incremental quick obtaining stress spectra,
It is characterized in that, the invalid stress amplitude of removal is to remove the small stress amplitude of some pairs of aging effects.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811324437.5A CN109408998B (en) | 2018-11-08 | 2018-11-08 | Fatigue life evaluation method by rapidly acquiring stress spectrum based on sample continuous increment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811324437.5A CN109408998B (en) | 2018-11-08 | 2018-11-08 | Fatigue life evaluation method by rapidly acquiring stress spectrum based on sample continuous increment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109408998A true CN109408998A (en) | 2019-03-01 |
CN109408998B CN109408998B (en) | 2023-01-03 |
Family
ID=65472537
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811324437.5A Active CN109408998B (en) | 2018-11-08 | 2018-11-08 | Fatigue life evaluation method by rapidly acquiring stress spectrum based on sample continuous increment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109408998B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111581798A (en) * | 2020-04-29 | 2020-08-25 | 中联重科股份有限公司 | Method and device for evaluating remaining life of support leg |
CN111686511A (en) * | 2020-06-28 | 2020-09-22 | 广州形银科技有限公司 | Sewage purification device for construction |
CN115358093A (en) * | 2022-10-18 | 2022-11-18 | 河南卫华重型机械股份有限公司 | Method for monitoring cracks of main beam of bridge crane in real time |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6480792B1 (en) * | 1998-10-22 | 2002-11-12 | British Aerospace Public Limited Company | Fatigue monitoring systems and methods incorporating neural networks |
US20060243055A1 (en) * | 2005-04-28 | 2006-11-02 | Sundermeyer Jeffry N | Systems and methods for determining fatigue life |
CN101408951A (en) * | 2008-11-18 | 2009-04-15 | 太原科技大学 | Method for obtaining equivalent load spectrum and estimating weariness residual longevity of bridge crane based on neural network |
CN102324040A (en) * | 2011-09-09 | 2012-01-18 | 太原科技大学 | Methods for acquiring equivalent load spectrum and estimating residual fatigue life of bridge crane |
CN105488328A (en) * | 2015-11-19 | 2016-04-13 | 北京航空航天大学 | Fatigue crack growth rate prediction method based on artificial neuron network |
CN105600682A (en) * | 2015-12-22 | 2016-05-25 | 江苏省特种设备安全监督检验研究院 | Crane strain monitoring method based on time-to-digital conversion technology |
CN108629092A (en) * | 2018-04-20 | 2018-10-09 | 北京航空航天大学 | One kind being based on the modified turbine disk subregion analysis method for reliability of dimensional effect |
-
2018
- 2018-11-08 CN CN201811324437.5A patent/CN109408998B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6480792B1 (en) * | 1998-10-22 | 2002-11-12 | British Aerospace Public Limited Company | Fatigue monitoring systems and methods incorporating neural networks |
US20060243055A1 (en) * | 2005-04-28 | 2006-11-02 | Sundermeyer Jeffry N | Systems and methods for determining fatigue life |
CN101408951A (en) * | 2008-11-18 | 2009-04-15 | 太原科技大学 | Method for obtaining equivalent load spectrum and estimating weariness residual longevity of bridge crane based on neural network |
CN102324040A (en) * | 2011-09-09 | 2012-01-18 | 太原科技大学 | Methods for acquiring equivalent load spectrum and estimating residual fatigue life of bridge crane |
CN105488328A (en) * | 2015-11-19 | 2016-04-13 | 北京航空航天大学 | Fatigue crack growth rate prediction method based on artificial neuron network |
CN105600682A (en) * | 2015-12-22 | 2016-05-25 | 江苏省特种设备安全监督检验研究院 | Crane strain monitoring method based on time-to-digital conversion technology |
CN108629092A (en) * | 2018-04-20 | 2018-10-09 | 北京航空航天大学 | One kind being based on the modified turbine disk subregion analysis method for reliability of dimensional effect |
Non-Patent Citations (5)
Title |
---|
JAMAL A.ABDALLA等: "Modeling and simulation of low-cycle fatigue life of steel reinforcing bars using artificial neural network", 《JOURNAL OFTHEFRANKLIN INSTITUTE》 * |
孟广伟 等: "基于径向基函数神经网络响应面法的装载机动臂疲劳可靠性", 《吉林大学学报》 * |
王爱红 等: "桥式起重机随机应力谱获取及疲劳剩余寿命估算", 《机械工程学报》 * |
范小宁 等: "基于人工神经网络获取起重机当量载荷谱的疲劳剩余寿命估算方法", 《机械工程学报》 * |
袁毅: "基于应力功率谱的结构振动疲劳寿命预测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111581798A (en) * | 2020-04-29 | 2020-08-25 | 中联重科股份有限公司 | Method and device for evaluating remaining life of support leg |
CN111581798B (en) * | 2020-04-29 | 2023-04-11 | 中联重科股份有限公司 | Method and device for evaluating the remaining life of a support leg |
CN111686511A (en) * | 2020-06-28 | 2020-09-22 | 广州形银科技有限公司 | Sewage purification device for construction |
CN115358093A (en) * | 2022-10-18 | 2022-11-18 | 河南卫华重型机械股份有限公司 | Method for monitoring cracks of main beam of bridge crane in real time |
CN115358093B (en) * | 2022-10-18 | 2023-02-28 | 河南卫华重型机械股份有限公司 | Method for monitoring cracks of main beam of bridge crane in real time |
Also Published As
Publication number | Publication date |
---|---|
CN109408998B (en) | 2023-01-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103544389B (en) | Autocrane method for diagnosing faults based on fault tree and fuzzy neural network | |
CN109408998A (en) | Estimating method for fatigue life is carried out based on sample incremental quick obtaining stress spectra | |
CN108108877A (en) | A kind of transmission line of electricity damage to crops caused by thunder methods of risk assessment based on BP neural network | |
CN111144039A (en) | Train dynamic weighing system and weighing method based on deep learning | |
CN101408951B (en) | Method for obtaining equivalent load spectrum and estimating weariness residual longevity of bridge crane based on neural network | |
CN104408562B (en) | A kind of photovoltaic system generating efficiency comprehensive estimation method based on BP neural network | |
CN109856515A (en) | A kind of direct current cables state of insulation judgment method and system | |
CN108763763A (en) | A kind of bridge structure strain-responsive abnormity early warning method | |
CN102323084A (en) | Progressive method for identifying damaged cable and support generalized displacement on the basis of mixed monitoring | |
CN203497948U (en) | Fuzzy neural network fault diagnosis device of auto crane | |
CN103278343B (en) | A kind of recognition methods of key structural fatigue component | |
CN107480787B (en) | Open-pit mine gyratory crusher fault diagnosis method based on BP neural network | |
CN106650959A (en) | Power distribution network repair ability assessment method based on improved grey clustering | |
CN105160496A (en) | Comprehensive evaluation method of enterprise electricity energy efficiency | |
CN107832789A (en) | Characteristic weighing k nearest neighbor method for diagnosing faults based on the conversion of average influence Value Data | |
CN104679655A (en) | Software system state evaluation method based on parameter correlation | |
CN112200263A (en) | Self-organizing federal clustering method applied to power distribution internet of things | |
CN104598970B (en) | Method for detecting work state of climbing frame group | |
CN107689015A (en) | A kind of improved power system bad data recognition method | |
CN107330264B (en) | Method for verifying reliability of bridge monitoring data | |
CN109978739A (en) | The adaptive learning method and computer system of knowledge based point Grasping level | |
CN109840891A (en) | A kind of intelligence strand tapered anchorage and prestressed monitoring method and detection system, terminal | |
CN105426665B (en) | Method is determined based on the DYNAMIC RELIABILITY of status monitoring | |
Li et al. | Reliability and sensitivity analysis of bridge crane structure | |
CN115330358A (en) | Education training management system and education training method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |